**Introduction **

Additive manufactured AM metallic materials are attractive materials because of minimum leading time and less materials loss, as components are formed directly from CAD data. The biggest problem on applications of AM metals is the fatigue strength, which is nearly half of the bulk metals [1-3]. One the of effective methods to improve fatigue strength is the application of mechanical surface treatments such as shot peening, cavitation peening [4, 5] and laser peening [6]. It has been reported that shot peening, cavitation peening and submerged laser peening improved the fatigue strength of AM metals [1-3].

In the case of as-built metals manufactured by powder bed melting AM method, surface roughness is considerably large due to un-melted particles (see Fig. 1). And also, when the depth of the surface defect of as-built specimen was measured by the fractured surface using SEM, it was about 200 μm for a direct metal laser sintering DMLS and an electron beam melting EBM [3]. Thus, in order to remove surface roughness and surface defects of AM metals with mechanical surface treatment, cavitation abrasive surface finishing (CASF) was proposed by the collaboration work with Tohoku University and Boeing, and it was demonstrated that CASF improved the fatigue strength of titanium alloy manufactured by EBM [7].

At CASF, a cavitating jet with abrasive was used for the mechanical surface treatment. The impact at cavitation bubble collapse, which is generated by the cavitating jet, introduces compressive residual stress- and work-hardening, and the abrasive, which is accelerated by the jet, is removed from the surface at the same time.

In the present paper, the enhancement of the fatigue strength of titanium alloy Ti6Al4V manufactured by EBM using CASF was demonstrated [7].

**Material and Methods **

The tested material was titanium alloy Ti6Al4V, and the fatigue specimens were manufactured by EBM. The thickness of the specimen was 2 ± 0.2 mm. The averaged diameter of the used particle size at EBM was about 75 μm. The spot size of the electron beam was 0.2 mm in diameter and the stacking pitch was 90 μm. After the EBM process, the specimens were heat-treated at 1208 K under vacuum for 105 minutes then cooled in argon gas. After that, aging was carried out at 978 K under vacuum for 2 hours, then cooled in argon gas.

The fatigue specimens were treated by a cavitating jet with abrasive. The details of CASF are in the reference [7]. The injection pressure was 62 MPa and the nozzle throat diameter for the jet was 0.64 mm. The specimen was placed in the recess and treated by moving the nozzle at constant speed v = 18 mm/s with the number of scan n. After each scan, the nozzle was moved 1.2 mm sideways. In this study, n was 1, 2, 3 and 4. As the length of the specimens was 90 mm, the processing time, tp, was 5 s, 10 s, 15 s and 20 s for n was 1, 2, 3 and 4 respectively. The specimens with and without CASF were tested by a conventional Schenk-type displacement-controlled plane bending fatigue tester at R = ―1. In order to find out the optimum processing time, the number of cycles to failure Nf at constant bending stress σa = 330 MPa was calculated as follows. As the displacement-controlled plane bending fatigue tester was used, the number of cycles at σa = 330 MPa was unknown. Thus, the number of cycles at σa ≈ 330 MPa of specimen was obtained experimentally, then Nf 330 was calculated from Nf at σa ≈ 330 MPa by the following procedure. It was assumed that the S-N curve for low cycle fatigue of non-treated specimens is described by Eq. (1), and that for treated specimens is described by Eq. (2), where, c1, c2 and c3 are constants. Thus, these S-N curves were parallel to each other.

(1)

(2)

Nf 330 for treated specimens was given by Eq. (3).

(3)

In the present experiment, c1 and c2were obtained from the 3 experimental data of non-treated specimens by the least-squares method, and c3 was obtained from c1 and the one experimental data for each treated specimen with σa ≈ 330 MPa respectively.

From Eq. (3), we get

(4)

From Eq. (4), we obtain Nf 330. In order to investigate the fatigue strength, the fatigue tests were carried out under the optimum processing time that maximizes the fatigue life, Nf 330. A test was considered a runout when a specimen exceeded 107 cycles and stopped.

As the plane bending fatigue strength was affected by the surface mechanical properties such as roughness, hardness and residuals stress, the surface roughness Rz was measured by a stylus type profilometer and the surface hardness HR15T was measured using a Rockwell superficial tester. The surface residuals stress was measured by a 2D-XRD method.

As the compressive residual stress σCR reduces applied stress, the fatigue life was improved as shown in Fig. 2. The a and b are constants. The surface hardness HR15T also improves fatigue life. On the other hand, the surface roughness Rz reduces fatigue life as shown in Fig. 2. The fatigue life Nf3 at σa can be estimated from σCR, HR15T and Rz by Eq. (5).

(5)

**Aspect of Specimen Surface Treated by CASF **

Figure 3 shows the aspect of specimen surface treated by CASF. At tp = 5 s, most unmelted particles were removed from the surface, but the wavy pattern was observed. At tp = 10 s, the wavy pattern became shallow, however, several surface defects were observed at the surface. The surface defects were removed from the surface at tp = 15 s, and the imbricate pattern became deeper at tp = 20 s.

In order to find out the optimum processing time, Fig. 4 reveals the number of cycles to failure Nf 330 at σa = 330 MPa changing with processing time. In Fig. 4, the surface hardness HR15T, the surface compressive residual stress σCR and the surface roughness Rz are also shown. The Nf 330, HR15T, σCR and Rz were normalized by the maximum values, 230,647 cycles, 92.2, 271.2 MPa and 108.8 μm respectively. The Nf 330 had a maximum at tp = 15 s. The HR15Tincreased with tp and was saturated, and the σCR increased with tp. The Rz decreased with tp and had a minimum at tp = 15 s, then increased at tp = 20 s. As mentioned, HR15T and σCR improved fatigue properties, and Rz decreased the fatigue life. This is why Nf 330 had a maximum at tp = 15 s.

**Improvement of Fatigue Strength by CASF**

Figure 5 illustrates the result of the plane-bending fatigue test to reveal the effect of CASF on the fatigue properties of as-built Ti6Al4V manufactured by EBM. The specimens were treated by CASF at tp = 15 s. As shown in Fig. 4, the fatigue life Nf 330 of as-built specimen was enhanced 2.46 times by CASF. When the fatigue strength at 107 was calculated by using Little’s method [8], it was 169 ± 8 MPa for as-built and 280 ± 10 MPa for CASF. Namely, CASF improved the fatigue strength at 107 about 66 % by CASF compared with the non-treated specimens.

In order to confirm the estimation method of fatigue life by Eq. (5), Fig. 6 shows the relationship between the experimental fatigue life Nf exp at σa ≈ 330 MPa and the estimated fatigue life Nf est at σa = 330 MPa. Both values were normalized by the number of cycles to failure of the specimen treated by CASF at tp = 15 s. The constants a and b in Eq. (5) were obtained by the least squares method using the five points of Fig. 4. When the correlation coefficient for the five data points was calculated, it was 0.958, meaning that the probability of non-correlation was less than 1.0%; thus, it can be concluded that the fatigue strength of AM Ti6Al4V treated by CASF was estimated from HR15T, σCR and Rz.

**Conclusions**

In order to demonstrate the enhancement of fatigue strength of additive manufactured AM metallic materials by cavitation abrasive surface finishing CASF, the titanium alloy Ti6Al4V manufactured by electron beam-melting EBM was treated by CASF and tested by a displacement-controlled plane fatigue test. It was revealed that the fatigue strength at 107 considering the surface roughness was improved 1.66 times by CASF compared with that of the as-built specimen.

**Acknowledgement **

This work was partly supported by JSPS KAKENHI Grant Number 17H03138, 18KK0103 and 20H02021.

**References**

[1] P. Edwards, A. O'Conner, and M. Ramulu, "Electron Beam Additive Manufacturing of Titanium Components: Properties and Performance," Journal of Manufacturing Science and Engineering, Trans. ASME, vol. 135, no. 6, paper no. 061016, pp. 1-7, 2013.

[2] H. Soyama, and Y. Okura, "The Use of Various Peening Methods to Improve the Fatigue Strength of Titanium Alloy Ti6Al4V Manufactured by Electron Beam Melting," AIMS Materials Science, vol. 5, no. 5, pp. 1000-1015, 2018.

[3] H. Soyama, and F. Takeo, "Effect of Various Peening Methods on the Fatigue Properties of Titanium Alloy Ti6Al4V Manufactured by Direct Metal Laser Sintering and Electron Beam Melting," Materials, vol. 13, no. 10, paper no. 2216, pp. 1-26, 2020.

[4] H. Soyama, "Key Factors and Applications of Cavitation Peening," International Journal of Peening Science and Technology, vol. 1, no. 1, pp. 3-60, 2017.

[5] H. Soyama, "Cavitation Peening: A Review," Metals, vol. 10, no. 2, paper no. 270, pp. 1-27, 2020.

[6] H. Soyama, "Comparison between the Improvements Made to the Fatigue Strength of Stainless Steel by Cavitation Peening, Water Jet Peening, Shot Peening and Laser Peening," Journal of Materials Processing Technology, vol. 269, pp. 65-78, 2019.

[7] H. Soyama, and D. Sanders, "Use of an Abrasive Water Cavitating Jet and Peening Process to Improve the Fatigue Strength of Titanium Alloy 6Al-4V Manufactured by the Electron Beam Powder Bed Melting (EBPB) Additive Manufacturing Method," JOM, vol. 71, no. 12, pp. 4311-4318, 2019.

[8] R. E. Little, "Estimating the Median Fatigue Limit for Very Small Up-and-Down Quantal Response Tests and for S-N Data with Runouts," ASTM STP, vol. 511, pp. 29-42, 1972.

Hitoshi Soyama (Ph.D. in Eng.) Professor

Department of Finemechanics

Tohoku University

6-6-01 Aoba, Aramaki, Aoba-ku, Sendai

980-8579, Japan

E-mail: soyama@mm.mech.tohoku.ac.jp

Daniel G. Sanders (Dr. Eng.)

Senior Technical Fellow

The Boeing Company, Seattle, WA, USA.

Affiliate Professor

The University of Washington

Seattle, WA, USA.